Prolonged jaundice (PJ) in healthy term neonates is common and frequently benign. It can, however, be the earliest manifestation of underlying liver disease. Its management requires a balanced approach, avoiding over-investigation of well babies while ensuring the early identification of those with pathology. Currently marked heterogeneity exists in the assessment of PJ. Over a two-year period we prospectively audited the management of PJ in two Level 3 neonatal units prior to and after the introduction of a rationalized investigation algorithm in keeping with the recently published British Society of Paediatric Gastroenterology, Hepatology and Nutrition (BSPGHAN) guidelines (i.e. clinical examination and stool inspection combined with measurement of split bilirubin). In this study we reviewed initial practice and then evaluated the impact of our change in practice. A total of 197 babies, 1.5% of live births, were referred with PJ. Of these, 105 babies were included in the first part of the study and 92 babies were included in the second part. No pathology relating to PJ, such as infection, hepatitis or liver disease, was identified. Following the introduction of our rationalized algorithm, we demonstrated a statistically significant reduction in the number of return appointments (28 versus 7; P < 0.0009) and repeat investigations (37 versus 7; P < 0.0001). This represented a saving of £1575-2625 per year in laboratory costs alone. Contemporaneously, three infants presented with biliary atresia, none of whom were identified by PJ screening and all of whom were over seven weeks old at diagnosis. A rationalized approach to the assessment of PJ reduces workload and is cost-effective; however, the limitations of selective screening, irrespective of how streamlined it is, remain--if babies are not identified and referred, they cannot be screened. Population-based methodologies offer an alternative approach to the identification of cholestatic liver disease and are worthy of further consideration.